According to more than N tutorials found on the Internet, it finally compiled and ran successfully. Record it here for the convenience of the next operation.
1. First of all, by default, you have downloaded VS2019 and installed desktop development using c++.
OpenCV version 4.5 download OpenCV download , Baidu network disk link: https://pan.baidu.com/s/1EgQwOlyK1mZj_hl9YsDHDA Extraction code: z2p1
Find the downloaded file, double-click to install (this is actually a decompression process)
Select the decompression directory, click Extract to automatically create a folder named opencv under the selected path. The decompressed content is as follows:
Next, add the following environment variables to the computer, Path in the system variables (need to be modified according to your own installation path):
My own is E:\opencv\build\x64\vc15\bin
After the environment variables are configured, restart the computer.
2. Configure OpenCV in VS2019
First open VS2019, create a new project, select C++, empty project.
Enter the project name and location of the project, click Create
Right click on the project name, select properties
Click the VC++ directory on the left , select X64 as the platform, and pay attention to the include directory and library directory on the right
First modify the containing directory, select it and click Edit
Click on the folder sign New and select Directory
Add two variables as shown in the figure below, click OK
E:\opencv\build\include
E:\opencv\build\include\opencv2
Add library directory (operation as above) add E:\opencv\build\x64\vc15\lib
Finally add Modify Additional Dependencies
Add a variable (manual input or paste)
opencv_world450d.lib , click OK (here I am using Debug, if using Release, change to opencv_world450.lib )
Finally, test whether the configuration is successful
Right click on the source file , Add -> New Item
Select the C++ file and enter the name main
Add the following code (just paste it directly)
#include<opencv2/opencv.hpp>
#include<iostream>
using namespace cv;
int main(int argc, char** argv) {
Mat image = imread("F:/1.jpg");
if (image.empty()) {
printf("could not load image...\n");
return -1;
}
//namedWindow("test_opencv_setup", 0);
imshow("test_opencv_srtup", image);
waitKey(0);
return 0;
}
Configure Cuda
Reference https://blog.csdn.net/fengxinzioo/article/details/109402921
Note: The ffmpeg and ippicv files that need to be downloaded when compiling versions above opencv4 are replaced into the .cache folder after downloading, rename and add the corresponding MD, and cmake will not have a warning again.
Update 2020-11-30
4.5
4.5 All cached files to be downloaded are uploaded to the network disk Link: link
Extraction code: n2kc
The principle is the same as above.
I found a lot of articles, only this one is simple, and the others are limited and I can't understand it.